[
https://issues.apache.org/jira/browse/IGNITE-27600?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vadim Pakhnushev updated IGNITE-27600:
--------------------------------------
Release Note: Fixed compute job failure after a node is restarted with
clean working directory.
> Compute job fails when node is restarted with clean working directory
> ---------------------------------------------------------------------
>
> Key: IGNITE-27600
> URL: https://issues.apache.org/jira/browse/IGNITE-27600
> Project: Ignite
> Issue Type: Bug
> Components: code deployment ai3
> Reporter: Vadim Pakhnushev
> Assignee: Vadim Pakhnushev
> Priority: Major
> Labels: ignite-3
> Time Spent: 10m
> Remaining Estimate: 0h
>
> When unit is deployed on a node and then node is restarted with clean working
> directory (keeping consistent id), compute requests on this node fail with
> the following exception:
> {noformat}
> org.apache.ignite.compute.ComputeException: IGN-COMPUTE-9 Job execution
> failed: java.lang.ClassNotFoundException:
> org.apache.ignite.internal.compute.ToStringJob. Deployment unit jobs:1.0.0
> doesn't exist
> {noformat}
> This happens because metastorage keeps node unit status DEPLOYED but the
> local directory doesn't contain the files anymore.
> We need to restore consistency on node start - either remove the node status,
> or use on-demand deploy to download the unit.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
